INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

It is well established that genotyping of Mycobacterium tuberculosis isolates provides valuable support for the investigation of suspected outbreaks, the detection of unsuspected transmission, the tracing of tuberculosis within a community, and the identification of possible sources of infection for newly diagnosed cases (1, 3, 9, 18). At the national or international level, fingerprinting allows strains from different geographic areas to be compared and the movement of individual strains to be tracked. IS6110 restriction fragment length polymorphism (RFLP) fingerprinting is the most widely used method for the genotyping of M. tuberculosis strains (19) because of its reproducibility and discriminatory power (9). Unfortunately, this method requires high-quality genomic DNA, which is not only difficult to prepare but also requires culturing of the organism, resulting in a long turnaround time. In addition, fingerprint interpretation and matching can be complicated and require sophisticated computer software for large-scale analysis.

In contrast, nucleic acid amplification-based assays do not require culturing of the organisms, allowing the analysis of samples in real time. In many PCR-based typing assays, the target DNA of interest is amplified and labeled by PCR and the labeled products are hybridized to an array of immobilized diagnostic probes. This method has been successfully used for the detection of mutations in drug resistance genes of M. tuberculosis (21) and for Mycobacterium species identification (17). Spoligotyping (6, 8, 12, 20), a reverse dot blot assay that detects the presence of a series of unique spacers in the direct repeat (DR) locus, meets the need for a simple and rapid method by which to distinguish M. tuberculosis complex strains. However, spoligotyping has significantly less discriminatory power than IS6110 RFLP fingerprinting for strains that contain more than five copies of IS6110 (9).

We report here a new method similar to spoligotyping that targets specific sites of insertion of IS6110. We refer to this insertion site mapping method as "insite" because it can be used to gain insight into the IS6110 insertion sites of an M. tuberculosis complex strain. Here we demonstrate that insite can be used to detect known IS6110 insertion sites to provide an IS6110-based fingerprint for M. tuberculosis complex strains.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains and clinical samples. M. tuberculosis strain H37Rv DNA and strain Erdman DNA were provided by J. Belisle (Colorado State University). M. tuberculosis strain CDC1551 and multidrug-resistant (MDR) strain W were from the Centers for Disease Control and Prevention mycobacteriology laboratory collection. A randomized, blinded set of 72 DNA samples from 60 M. tuberculosis clinical isolates prepared for an unrelated study of secondary typing was used to evaluate reproducibility. Genomic DNA was prepared using standard methods (19).

LMPCR amplification. Labeled target DNAs were prepared by a modification of previously published mixed-linker PCR (7) and linker-mediated PCR (LMPCR) methods using asymmetric linkers (11, 15). The basis for specificity of the amplification is described in the original paper (7). A partially double-stranded linker was prepared by combining oligonucleotides linker 1 and linker 2 (Table 1) at a final concentration of 20 µM each in 1× PCR Buffer (Perkin-Elmer Applied Biosystems, Foster City, Calif.). The mixture was incubated at 80°C for 5 min and then slowly cooled to 25°C. The annealed linker was stored at 20°C. Genomic DNA (0.1 to 10 ng) was digested with 20 U of HhaI in 20 µl of 1× Reaction Buffer 4 (NEB, Beverly, Mass.) for 3 h at 37°C. The linker (20 pmol) was ligated to 10 µl of digested DNA with 1 U of T4 DNA ligase (Gibco-BRL) in 20 µl of 1× Ligation Buffer overnight at 16°C. The ligation products were digested with 20 U of HhaI for 30 min at 37°C to eliminate any religated HhaI sites. Templates for the primary PCR amplification were prepared by diluting the sample 1:100 in 10 mM Tris-HCl, pH 8.3. The primary PCR amplification (50-µl total volume) consisted of 1× PCR Buffer, 1 U of AmpliTaq DNA polymerase (Perkin-Elmer Applied Biosystems), deoxynucleoside triphosphates (0.2 mM each), 5% dimethyl sulfoxide, primer IS-Primary (1 µM), linker primer (1 µM), and 5 µl of the diluted template. Amplification was performed in a GeneAmp PCR System 9700 (Perkin-Elmer Applied Biosystems) with an amplification profile that consisted of an initial denaturation step of 94°C for 1 min; 40 cycles of denaturation at 94°C for 30 s, annealing at 65°C for 30 s, and extension at 72°C for 1 min; and a final extension at 72°C for 1 min. Templates for the nested PCR (labeling reaction) were prepared by diluting the primary PCR products 1:100 in 10 mM Tris-HCl, pH 8.3. The nested PCR conditions were identical to the primary PCR conditions, except that the primer IS-Primary was replaced with the biotinylated primer IS-Nested. All primer sequences, including those used for the specific amplification of the IS6110 flanking sequences in strain 1551, are listed in Table 1.

Membrane preparation. Amino-linked oligonucleotides (Table 2) were synthesized with a C₆ spacer by the Biotechnology Core Facility, Centers for Disease Control and Prevention, and were diluted to the indicated concentration with 0.5 M NaHCO₃. Membranes were prepared following previously published procedures (12). A Biodyne C membrane (Pall Biosupport, Ann Arbor, Mich.) was activated by incubation in 16% (wt/vol) 1-ethyl-2-(3-dimethylaminopropyl)carbodiimide (Sigma Chemical, St. Louis, Mo.) for 10 min at 25°C. Following a brief wash with deionized water, 150 µl of each diluted oligonucleotide was applied in a line by using a miniblotter system (MN45; Immunetics, Cambridge, Mass.). After a 5-min incubation at room temperature, the oligonucleotide solutions were removed from the membrane by aspiration. The membrane was inactivated by incubation in 100 mM NaOH for 9 min at room temperature, followed by a brief wash with 2× SSPE (0.36 M NaCl, 20 mM NaH₂PO₄, 2 mM EDTA, pH 7.7; Gibco-BRL, Grand Island, N.Y.) and a 5-min incubation in 2× SSPE-0.1% sodium dodecyl sulfate (SDS) at 58°C. The membrane was incubated in 20 mM EDTA for 20 min at room temperature and stored at 4°C.

Insite assay. A 20-µl volume of LMPCR product was diluted with 150 µl of 2× SSPE-0.1% SDS and boiled for 10 min. The samples were then cooled on ice for 2 min, briefly centrifuged, and stored on ice until use. The insite membrane was incubated in 2× SSPE-0.1% SDS for 5 min at room temperature and then inserted into the miniblotter apparatus such that the lines of previously applied oligonucleotides were perpendicular to the sample lanes. Residual liquid was removed by vacuum aspiration. For each sample, 150 µl of diluted PCR-amplified biotinylated product was added to one slot of the miniblotter. The entire apparatus was then incubated at 58°C for 1 h. Following hybridization, the samples were removed by vacuum aspiration for at least 1 min. The membrane was removed from the miniblotter and washed twice with 25 ml of 2× SSPE-0.5% SDS for 10 min at 58°C. The membrane was transferred to a roller bottle and incubated with 3.75 U of streptavidin-POD (Boehringer Mannheim, Indianapolis, Ind.) diluted in 30 ml of 2× SSPE-0.5% SDS for 1 h at 39°C. The membrane was washed twice with 25 ml of 2× SSPE-0.5% SDS for 10 min at 39°C and then twice with 100 ml of 2× SSPE for 5 min at room temperature. ECL chemiluminescence detection reagents (26 ml/membrane; Amersham Pharmacia Biotech, Piscataway, N.J.) were added, and the membrane was exposed to X-Omat AR film (Kodak) for 30 min to 16 h. For repeated use (up to 12 times), the membrane was stripped by a 1-h incubation in 1% SDS at 76°C. The membrane was incubated in 20 mM EDTA for 20 min at room temperature and stored at 4°C.

Identification of strain Erdman insertion sites. LMPCR products prepared from strain Erdman genomic DNA were separated in a 2% agarose gel. Individual amplicons were isolated from the gel using the QIAquick Gel Extraction Kit (Qiagen, Valencia, Calif.) and inserted into the cloning vector pT7Blue using the Perfectly Blunt Cloning Kit (Novagen, Madison, Wis.). The inserts were sequenced using the ABI Prism BigDye Terminator Cycle Sequencing Ready Reaction Kit (Perkin-Elmer Applied Biosystems) and ABI 373 sequencer with XL upgrade.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Reverse dot blot hybridization assays involve the amplification and labeling of the DNA sequence of interest (the target DNA), followed by hybridization of the labeled amplicon to oligonucleotides (the probes) immobilized on a membrane. For example, in spoligotyping (6, 8, 12, 20), the individual spacers in the DR locus serve as target DNA and are amplified using biotin-labeled primers specific for the 36-bp DR element. The labeled target DNAs are hybridized to 43 immobilized oligonucleotide probes, each representing a unique spacer. M. tuberculosis complex strains can be typed based on the presence or absence of spacers. We desired to develop a reverse dot blot method by which to detect specific IS6110 insertion sites. The method is based on the premise that, under the appropriate hybridization conditions, a probe representing the sequence at the junction of the IS6110 insertion site would hybridize specifically with the corresponding target DNA but would not hybridize with either IS6110 or the flanking DNA alone. A set of labeled target DNAs representing the various copies of IS6110 and their flanking sequences is prepared by LMPCR amplification (Fig. 1) using a biotinylated primer. The oligonucleotide probes correspond to known IS6110 insertion sites and are complementary to the biotin-labeled strand generated during LMPCR amplification of the target DNA. The target DNA is hybridized to the immobilized probes, and the presence or absence of a copy of IS6110 in a particular insertion site in the genome can be detected.

View larger version (18K): [in this window] [in a new window]   FIG. 1.   General strategy of the insite assay. IS6110-specific LMPCR amplification is used to generate a set of labeled amplicons representing the copies of IS6110 and their flanking sequences. The biotinylated IS-Nested primer is indicated. Oligonucleotide probes are designed to consist of both the right end of IS6110 and the flanking sequence and are complementary to the biotinylated strand generated during LMPCR amplification. Probes 1 to 5 were designed to detect the first insertion site in M. tuberculosis strain 1551-E. The IS6110 sequence is indicated by lowercase letters, the flanking sequence is indicated by capital letters, and the nucleotides involved in predicted secondary structures are underlined.

The potential use of this assay was investigated using the four IS6110 insertion sites in M. tuberculosis strain CDC1551. First, to investigate the optimal composition of the oligonucleotide probe, several different probes for the detection of one of the insertions sites, 1551-E, were designed (Fig. 1). Probes 1551-E(10/10), 1551-E(12/8), and 1551-E(15/7) are 20- to 22-base oligonucleotide probes that contain slightly different flanking sequence and IS6110 sequence lengths. Oligonucleotide probes 1551-1(+3) and 1551-1(3) contain a three-base insertion and a three-base deletion, respectively, located between the end of IS6110 and the flanking sequence, simulating insertion of IS6110 at three bases on either side of the actual 1551-E insertion site. Each probe (25 pmol) was chemically cross-linked to an activated Biodyne-C membrane. To generate labeled target DNA for this copy of IS6110, a primer corresponding to a sequence 20 bases downstream of insertion site 1551-E and the biotinylated IS-Nested primer were used, producing an amplicon analogous to the fragment generated during LMPCR amplification. The target DNA bound to the five immobilized probes with varying efficiency (Fig. 2). Probe 1551-E(10/10) produced the strongest signal intensity, and probe 1551-E(12/8) produced the weakest. The target DNA did not hybridize to the probe containing either the three-base insertion or the three-base deletion. This suggested that 20- to 22-base oligonucleotides are of sufficient length for hybridization under the conditions used and that the probe must conform closely to the IS6110 insertion site in order to bind the target DNA.

View larger version (13K): [in this window] [in a new window]   FIG. 2.   Specificity of strain CDC1551 insertion site probes. Oligonucleotide probes designed to detect the four insertion sites in M. tuberculosis strain CDC1551 were hybridized with biotinylated PCR products. Lanes: 1, PCR-amplified IS6110 flank E; 2, PCR-amplified IS6110 flank 2; 3, PCR-amplified IS6110 flank 3; 4, PCR-amplified IS6110 flank 4; 5, LMPCR products for M. tuberculosis strain CDC1551.

To test the specificity of the assay, probes for the three remaining insertion sites in strain CDC1551 (1551-2, 1551-3, and 1551-4, Table 2) were designed with 10 bases of the IS6110 sequence and 10 bases of the flanking sequence. Each of these probes (125 pmol), along with each of the five probes for 1551-E (25 pmol), was cross-linked to a membrane. Each of the four insertion sites was amplified individually using specific primers and the IS-Nested primer. Each target hybridized specifically to the corresponding probe, although differences in signal intensity were again observed (Fig. 2). In particular, the signal for probe 1551-2 was barely detectable. Because the same amount of each target was used, this suggested that the probes were binding with different efficiencies. Analysis of the probe nucleotide sequences revealed that both probes 1551-2 and 1551-E(12/8) were capable of forming strong secondary structures whereas probes 1551-E(10/10) and 1551-E(15/7) were not. To test the hypothesis that secondary structures reduce the efficiency of target binding, a second probe, 1551-2a, was designed so that the hairpin structure was destroyed. As can be seen in Fig. 2, probe 1551-2a bound the target DNA significantly better than did probe 1551-2.

To determine if the assay could detect all four insertion sites that had been amplified simultaneously, the membrane containing the nine probes described above was incubated with LMPCR products from CDC1551 (Fig. 2). Each insertion site could be detected successfully but not equally. During LMPCR amplification, individual fragments may not be amplified with equal efficiency, and in the final set of amplicons, smaller fragments are often represented far more than are larger fragments. The unequal proportions of each target in the LMPCR amplification products can be seen after agarose gel electrophoresis analysis (data not shown) and in the variation of signal intensity after hybridization with the insertion site probes (Fig. 2).

Next, the insite assay was expanded to include selected insertion sites for strains H37Rv and Erdman and MDR strain W (Table 2). Oligonucleotide probes designed to detect insertion sites for 13 of the 16 copies of IS6110 in the H37Rv genome (accession number AL123456) were added to the membrane. Rv1 and Rv14 are identical, and one probe (Rv1) detects both. A probe meant to detect one of the remaining insertion sites was already included; Rv11 is identical to 1551-4. Probes for the final two copies of IS6110 were not included for technical reasons. Rv-IS6110 3 was not present in our stock of H37Rv, as determined by PCR amplification of this region (data not shown). Rv-IS6110 7 is inserted two bases upstream of an HhaI site, resulting in an only five-base flanking sequence in the linker-mediated amplicon of this copy, which is insufficient for specific hybridization. Also included on the insite membrane were oligonucleotides designed to detect three of the IS6110 insertion sites present in MDR strain W (11, 14), seven of the eight IS6110 insertion sites in strain Erdman as sequenced in this study (E1, E3, E6 to -8, DK5, and 1551-1), and seven previously published IS6110 insertion sites (DK4 to -6, 9, 10, 13, and 14; reference 5). The quantity of oligonucleotide probe immobilized on the insite membrane was determined empirically to obtain signals of similar intensity for each insertion site. Amounts loaded ranged from 6.25 to 250 pmol (Table 2).

The results of the hybridization of LMPCR products of strains 1551, Erdman, H37Rv, and W can be seen in Fig. 3. With one exception, each labeled product tested hybridized specifically with oligonucleotide probes containing the respective insertion site and gave the expected insertion site, or insite, profile. The LMPCR product from strain H37Rv consistently hybridized with probe DK4 (Fig. 3, lane 3), which differs from Rv10 in three positions. The 3' 14 bases in Rv10 and DK4 are perfectly matched and the calculated melting temperature (T_m) of this 14-mer oligonucleotide is 62.5°C, above the hybridization temperature used. One insertion site, Rv12, was not detected by the probes used in this assay, and two additional insertion sites, MDR strain W B2 and Rv16, were detected only weakly. The presence of Rv12 and Rv16 in the H37Rv strain used was confirmed by PCR using site-specific primers, and the insertion sites in these PCR fragments could easily be detected after hybridization with the probes (data not shown), suggesting that the inability to detect the insertion site after hybridization with the LMPCR products was due to insufficient labeled target in the set of amplicons.

View larger version (47K): [in this window] [in a new window]   FIG. 3.   Evaluation of insertion site probes using strains with defined insertion sites. LMPCR products were hybridized to the immobilized oligonucleotide probes (Oligos) listed in Table 2. Lanes: 1, MDR M. tuberculosis strain W; 2, strain H37Rv; 3, strain Erdman; 4, strain CDC1551.

An oligonucleotide probe meant to detect spacer 35 in the DR locus (20) was included on the insite membrane. Note that the spacer numbers used here refer to the recent study on the evolution of the DR locus by van Embden et al. (20) and not to the spacer numbers originally used in the description of the spoligotyping method (8); new spacer 35 is previous spacer 24. Most M. tuberculosis strains, including strains CDC1551 and H37Rv, have an IS6110 insertion in the DR preceding spacer 35, resulting in the amplification of this spacer during LMPCR amplification for insite analysis. As can be seen in Fig. 3, insite samples prepared for these strains hybridized to spacer 35. Strains Erdman and W also have copies of IS6110 in the DR locus. Strain Erdman has an asymmetric insertion in the DR preceding the second copy of spacer 35, followed by spacers 42 and 51, resulting in the hybridization of insite samples to both insite probe E6 (DR locus insertion) and spacer 35. Spacer 35 is not present in the DR locus of strain W (20).

To test the reproducibility of the assay, LMPCR products from a randomized set of 72 DNA samples (Table 3) from 60 M. tuberculosis isolates were hybridized to the membrane containing the 32 insertion site probes (Fig. 4). The set included duplicate samples from 12 isolates with IS6110 copy numbers ranging from 0 to 20 and 48 isolates from tuberculosis investigations in Arkansas and Tennessee. Although the insertion site profiles of the 12 duplicate samples were not particularly informative, they were reproducible.

View larger version (66K): [in this window] [in a new window]   FIG. 4.   Insite assay of a randomized, blinded set of M. tuberculosis DNA samples from clinical isolates as described in Table 3. The oligonucleotide probes (oligos) are as ordered in Table 2.

Thirty isolates from Arkansas have been grouped into six clusters. The 11 isolates in clusters A to C were from patients with epidemiologic links and match by IS6110 RFLP fingerprints and spoligotype patterns (data not shown). All of the samples within each cluster had identical insite patterns. Cluster A (IS6110 copy no. 5) was characterized as having the four insertion sites in strain CDC1551. For cluster B (IS6110 copy no., 10), only insertion site 1551-4 was detected, and for cluster C (IS6110 copy no. 12), insertion sites Rv6 and 1551-4 were detected. Spacer 35 was detected for cluster B but not for cluster C, which agrees with its respective spoligotype pattern (data not shown).

The 19 isolates in clusters D to F are from patients with no epidemiologic links, but the isolates match by IS6110 RFLP. Cluster D contains two isolates with a 13-band IS6110 RFLP fingerprint and identical spoligotype patterns; the insite profiles (Rv1, Rv2, Rv5, Rv8, Rv9, Rv10, Rv13, 1551-3, 1551-4) of the two isolates were similar, with the addition of the E7 insertion being identified for one of the samples. With a single exception, cluster E was characterized as containing only the 1551-E and 1551-4 insertion sites. The insite profile of sample 4 (E1, E3, E6, DK5, 1551-E) was similar to that of strain Erdman; the spoligotype pattern of this strain also differed from others in the cluster (data not shown), suggesting laboratory error in the handling of this culture. Cluster F also contained a single sample whose insite profile differed from the other five in the cluster. Although the cluster was characterized as containing the four insertion sites in strain 1551, strain 28 contained only the A1 insertion site, which is typically found in Beijing family strains (10). This result is in agreement with its nine-spot spoligotype pattern (data not shown) characteristic of Beijing family strains (16), again suggesting laboratory error.

Twelve DNA samples were from isolates associated with a tuberculosis outbreak in Tennessee; all of the isolates have matching IS6110 RFLP fingerprints and spoligotype patterns and are from patients with epidemiologic links. These isolates have been identified as M. tuberculosis strain CDC1551, which can also be identified by the 1551 PCR assay (14). Five additional isolates have an IS6110 RFLP fingerprint identical to that of strain 1551 but are unrelated to the outbreak, have two different spoligotype patterns, and are negative in the 1551 PCR assay (data not shown). A final isolate is characterized as having a positive 1551 diagnostic PCR result but has a different four-band IS6110 RFLP pattern and a different spoligotype pattern. All 18 of the isolates were shown to contain the four strain 1551 insertion sites. LMPCR products for the last five samples and one of the CDC1551 outbreak samples were prepared from a glass bead-mickle apparatus crude lysate of each, which demonstrates that it is not necessary to prepare purified genomic DNA samples for the insite assay.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

We developed a method, designated insite, which can be used to detect IS6110 insertion sites. The method relies on the hybridization of biotin-labeled target DNA, generated by IS6110-specific LMPCR amplification, to oligonucleotide probes representing known sites of insertion immobilized on a membrane. The assay can easily be completed in 3 days, and because the method is amplification based, it does not require large-scale purification of high-quality genomic DNA. A single colony suspension is a sufficient template for LMPCR amplification (2). The specificity of the assay was demonstrated by both the specific hybridization of individual PCR products and the hybridization patterns of LMPCR amplification products from strains with well-defined insertion sites. The reproducibility of the assay was demonstrated by the identical insite profiles generated for 12 matched pairs of strains and for samples from tuberculosis outbreak investigation clusters.

Some insertion sites were detected more reliably than others, and a combination of factors are likely to be responsible for this effect. Amplicons representing each copy of IS6110 are not represented equally in the hybridizing DNA sample because all targets are not amplified with equal efficiency using the LMPCR method. In particular, larger amplicons are not amplified as efficiently as smaller ones. The composition of the oligonucleotide probe also influences hybridization efficiency. Oligonucleotide probes that contain strong secondary structure hybridize with the DNA sample with reduced efficiency under our assay conditions. These difficulties may be overcome by decreasing the length of the IS6110 sequence (which has a very high GC content) in the probe to reduce the secondary structure and by increasing the amount of oligonucleotide immobilized on the membrane. For probes with high T_m values (>70°C), reducing the length of the IS6110 sequence should also increase the specificity. The hybridization of Rv-IS6110 10 target DNA to probe DK4 highlights the importance of considering T_m values during oligonucleotide probe design.

The majority of the IS6110 insertion sites in this study were chosen simply because they are in strains with previously sequenced insertion sites. The four strains used in the development of this method contain a wide variation in IS6110 copy number (4, 8, 16, or 22 copies) and share relatively few insertion sites. It was not expected that the insertion sites chosen for method development would be of great use in discriminating between strains of M. tuberculosis. Although the insertion site profiles of the blinded set of 60 isolates showed the reproducibility of the assay, limited discrimination of strains was observed. It should also be noted that three of the insertion sites most often detected in these 60 isolates are also found in most low-copy-number strains. Fomukong et al. (5) showed that of 42 strains with four or five copies of IS6110 (all with different RFLP patterns), 88% had the 1551-1 (DK1) insertion, 76% had the 1551-3 (DK3) insertion, and 97% had the 1551-4 (DK2) insertion. Because over half of our set of 60 isolates have fewer than five copies of IS6110, it is not surprising that only a limited number of insertion sites were repeatedly detected.

By using a carefully selected, expanded set of insertion sites, the insite method should be useful for M. tuberculosis outbreak investigations. Specimens with clearly different insite patterns could be assumed to be different, while specimens with the same insite pattern could be further analyzed using a second fingerprinting method. This "carefully selected set of insertion sites" will have to be determined empirically by screening a large number of isolates using the insite method itself along with sequencing of LMPCR products. The three probes for strain W were included in this study as examples of carefully selected insertion sites. As described by Kurepina et al. (10), the evolution of MDR strain W can be followed using IS6110 transposition events. The A1 insertion has only been identified in group 1 isolates, the B2 insertion can be found in a subset of group 1 isolates (e.g., strain W, strain N), and the B1 insertion, which presumably occurred after the acquisition of the MDR phenotype, can only be found in MDR strain W. Kremer et al. (9) showed that genotyping can be used successfully to define genotype families within the M. tuberculosis complex family. The insite approach will also be useful for defining large families of strains. A small set of conserved insertion sites will likely define each family, with additional key sites defining major branches. The insite assay will recognize these insertion sites even when mutations have altered the size of the PvuII fragments for the conserved sites and the overall IS6110 patterns have diverged significantly.

In most instances, this method will provide less discrimination than that seen with traditional IS6110 RFLP fingerprinting because only known insertion sites can be detected. An important exception to this will be the detection of IS6110 in hot spotsvery small regions of the genome in which a large number of IS6110 insertion sites have been observed. Although these produce very small variations in fragment length that are often indistinguishable using RFLP typing methods, the insite assay should be able to easily discriminate between them. For example, one of the hot spots for IS6110 in the M. tuberculosis chromosome is the ipl locus (4). Three probes (Rv1, E6, and E7) specific for IS6110 insertion sites in this locus were among the 32 probes utilized in this study, and all were highly specific for their respective insertion site sequences.

Although the insite method is limited to known insertion sites, it is a significant improvement over currently used methods for determining the presence or absence of IS6110 in a specific position within the genome. In contrast to either probing IS6110 RFLP membranes with individual probes or amplifying a flanking sequence using specific primers, the insite method using a line blot apparatus can simultaneously screen up to 45 different samples for 45 different insertion sites. Use of the new microarray technologies should allow the screening of almost an unlimited number of probes. In conclusion, we have developed a method, insite, which is capable of the specific detection of IS6110 insertion sites at the nucleotide level. The application of the assay for M. tuberculosis strain differentiation will require additional knowledge regarding the distribution of IS6110 insertion sites in genotype families.